Theranostics 2020, Vol. 10, Issue 21 9458

Ivyspring International Publisher Theranostics 2020; 10(21): 9458-9476. doi: 10.7150/thno.44688 Research Paper Loss of SIRT4 promotes the self-renewal of Breast Cancer Stem Cells Lutao Du1, Xiaoyan Liu1, Yidan Ren1, Juan Li1, Peilong Li1, Qinlian Jiao1, 2, Peng Meng3, Fang Wang4, Yuli Wang1, Yun-shan Wang1 and Chuanxin Wang1

1. Department of Clinical Laboratory, The Second Hospital, Cheeloo College of Medicine, Shandong University, 247 Beiyuan Street, Jinan, Shandong, 250033, China. 2. International Biotechnology R&D Center, Shandong University School of Ocean, 180 Wenhua Xi Road, Weihai, Shandong 264209, China. 3. The Medical Department of IVD Division, 3D Medicines, Inc., Pujiang Hi‑tech Park, Shanghai 201114, China. 4. Institute of basic medicine, The Second Hospital, Cheeloo College of Medicine, Shandong University, 247 Beiyuan Street, Jinan, Shandong, 250033, China.

 Corresponding authors: Yun-shan Wang, Department of Clinical Laboratory, The Second Hospital, Cheeloo College of Medicine, Shandong University, 247 Beiyuan Street, Jinan, Shandong, 250033, China. E-mail: [email protected]; Phone Number: 86-531-85875118; Chuanxin Wang, Department of Clinical Laboratory, The Second Hospital, Cheeloo College of Medicine, Shandong University, 247 Beiyuan Street, Jinan, Shandong, 250033, China. E-mail: [email protected]; Phone Number: 86-531-85875118.

© The author(s). This is an open access article distributed under the terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0/). See http://ivyspring.com/terms for full terms and conditions.

Received: 2020.02.08; Accepted: 2020.07.07; Published: 2020.07.25

Abstract Rationale: It has been proposed that cancer stem/progenitor cells (or tumor-initiating cells, TICs) account for breast cancer initiation and progression. are adenine dinucleotide (NAD+)-dependent class-III histone deacetylases and mediate various basic biological processes, including metabolic homeostasis. However, interplay and cross-regulation among the family are not fully understood. As one of the least studied sirtuin family members, the mitochondrial sirtuin SIRT4 is a tumor suppressor in various cancers. However, its role in cancer stemness, as well as initiation and progression of breast cancer, remains unknown. Methods: The expression of SIRT4 in breast cancer was analyzed using the TCGA breast cancer database and 3 GSEA data. Normal breast epithelial cells MCF10A and breast cancer cell lines MCF-7, MDA-MB-231, BT549, MDA-MB-468 were used to establish SIRT4 gene knockdown and corresponding overexpression cells. Identified MTT cytotoxicity assays, cell invasion and motility assay, sorting of SP, confocal immunofluorescence microscopy, mouse mammary stem cell analysis, and glucose production, clonogenic and sphere-formation assay, mass spectrometric metabolomics analysis and ChIP-seq to further explore SIRT4 biological role in breast cancer. Results: We elucidated a novel role for SIRT4 in the negative regulation of mammary gland development and stemness, which is related to the mammary tumorigenesis. We also uncovered an inverse correlation between SIRT4 and SIRT1. Most importantly, SIRT4 negatively regulates SIRT1 expression via repressing glutamine metabolism. Besides, we identified H4K16ac and BRCA1 as new prime targets of SIRT4 in breast cancer. Conclusions: These results demonstrate that SIRT4 exerts its tumor-suppressive activity via modulating SIRT1 expression in breast cancer and provide a novel cross-talk between mitochondrial and nuclear sirtuins.

Key words: SIRT4, SIRT1, breast cancer, cancer stemness, glutamine metabolism

Introduction Sirtuins (SIRTs) are nicotinamide adenine stress-responsive that direct various post- dinucleotide (NAD+)-dependent class-III histone translational modifications of histones and down- deacetylases (HDACs) that belong to the silent stream non-histone targets, the seven mammalian information regulator 2 (SIR2) family. As the sirtuins (SIRT1–7) are considered to be master

http://www.thno.org Theranostics 2020, Vol. 10, Issue 21 9459 regulators of several fundamental biological This study is to identify a mechanism linking processes, including cell death and survival, decreased SIRT4 expression to breast tumor-initiating remodeling, energy metabolism, aging, cells (BTICs) regulation and cancer progression. Here, , and tumor development [1-3]. we show that SIRT4 possesses its tumor-suppressive Among the sirtuin family, SIRT1 is the founding effect on breast cancer by inhibiting glutamine and the most extensively studied member. SIRT1 is metabolism and thereby affecting the levels of ubiquitously expressed in the nucleus but shuttles SIRT1, which modulates the patterns of between the nucleus and cytoplasm [4]. Similar to histones H4 and regulates stemness via BRCA1. other HDACs, aberrant SIRT1 expression is reported in multiple human malignancies, namely, gastric, Materials and Methods colon, prostate, and ovarian cancers, as well as breast Chemicals and Antibodies cancer [5-11]. The implications of SIRT1 in breast cancer occurrence and progression have been Lipofectamine 2000 transfection and TRIZOL LS investigated over recent years. However, the exact reagents were bought from Invitrogen (Grand Island, role is still controversial due to its contradictory NY, USA). The DAB substrate kit has been purchased functional roles. from Vector Laboratories, Inc (Burlingame, CA, USA). Unlike SIRT1-3 and SIRT7, which are primarily Abcam (Cambridge, MA, USA) provided antibodies deacetylases, SIRT4 serves as both an ADP- towards SIRT4 (#: ab10140), SIRT1 (#: ab189494), ribosyl- and lysine deacetylase. It is a H4K16ac (#: ab109463), H3K9ac (#: ab272105), Oct4 relatively unstudied member of the SIRT family [12]. (#: ab19857), SOX2 (#: ab97959). Nanog (#: 8822), SIRT4 is located within mitochondria and regulates BRCA1 (#: 9010) and β-actin (#: 4970) antibodies were catabolism of multiple nutrients [13]. Despite the obtained from Cell Signaling Technology (Danvers, regulatory function in metabolism, SIRT4 was found MA, USA). Unless, in any other case noted, all other to be involved in mediating cell cycle progression and used chemicals were from Sigma (St. Louis, MO, signaling pathways like MEK/ERK or AMPK/ USA). mTORC [14-16]. So far, emerging evidence on the role Cell lines and cell culture of SIRT4 as a has been shown. Normal mammary epithelial cell line MCF10A According to The Cancer Genome Atlas (TCGA) and breast cancer cell lines MCF-7, MDA-MB-231, database analysis, SIRT4 mRNA levels in human BT549, MDA-MB-468 and HEK 293 Phoenix ampho gastric, breast, bladder, colon, thyroid, and ovarian cells were purchased respectively from the Cell Bank cancers were decreased compared with normal tissues of different types culture collections of Shanghai [17, 18]. Decreased SIRT4 protein levels were Institute of Cell Biology and Chinese Academy of associated with poor prognosis in colon, lung, and Sciences. As earlier mentioned, cell lines were esophageal cancer [16, 19, 20]. Intriguingly, although regularly cultured. Typically, cell lines were kept at 37 more tumor tissue samples were found SIRT4 positive °C in an environment comprising 5% CO2 in than adjacent nontumor tissue samples by immuno- Dulbecco's modified Eagle medium or RPMI 1640 histochemical staining, low SIRT4 expression was (Gibco) supplemented with 10% fetal bovine serum. associated with poor overall survival in breast cancers Cells were passed every 1-2 days to keep logarithmic patients, especially in Luminal A patients [21]. growth. All these cell lines were regularly screened However, critical mechanisms linking reduced SIRT4 for mycoplasma contamination by using the expression to breast cancer progression remain Universal mycoplasma detection kit, which was unclear. It has therefore been proposed that loss of obtained from ATCC (Manassas, VA), and the last BRCA1 function in basal stem cells results in tumor mycoplasma test was conducted in May 2017. formation associated with a block in luminal Mycoplasma-free cell lines have been used in all our differentiation [22]. experiments. In contrast, recent work showed an increase in luminal progenitor numbers in the breast tissue of Patients and specimens BRCA1 mutation carriers and a correlation between Randomly, samples were gathered from patients the profile of normal human luminal with breast cancer who underwent curative resection progenitors and basal-like breast cancers. Compared with informed consent between 2010 and 2015 at Qilu with most sporadic breast cancers, those arising in Hospital, Shandong University. All tissues were carriers of BRCA1 mutations usually have distinctive collected immediately after tumor resection, snap- pathological characteristics. A study suggests that a frozen into liquid nitrogen, and then stored at -80 °C. role for BRCA1 in the determination of stem-cell fate Follow-up information was summarized at the end of may explain this phenomenon [23]. December 2017 with an average observation interval

http://www.thno.org Theranostics 2020, Vol. 10, Issue 21 9460 of 75.4 months. Study protocols were endorsed by with a Coherent Enterprise II laser-emitting MLUV. Shandong University's Hospital Ethics Committee. The Hoechst 33342 emission was first divided using a Based on the Helsinki declaration, written informed short-pass 610-nm dichroic filter and the red and blue consent was obtained from patients. emissions were obtained through 670/30 and 450/65 nm bandpass filters, respectively. MTT Cytotoxicity assays A total of 2,000 cells/well were plated 12 hours Quantitative RT-PCR before treatment in 96 well plates. The cells were then Total RNA extraction has been done by using handled for 72 hrs with specified agents. 10% v/v of 5 Trizol reagent (Invitrogen), and cDNA was obtained mg/ml 3-[4,5-dimethylthiazol-2-yl]-2,5-diphenyl- by using SuperScript II Reverse Transcriptase tetrazolium bromide (MTT) solution was introduced purchased from Invitrogen. The ABI PRISM 7900HT for a duration of 2 hr. The medium was then sequence detection system performed the qRT-PCR, discarded, and the cells dissolved in DMSO (Sigma, St and it has been involved in data collection during the Louis, MO). Relative Cytotoxicity was determined by experiments. using a BMG Labtek plate reader to measure the absorbance at 570 nm. All expected experiments were Immunoblotting conducted in triplicates, and the mean was calculated For immunoblotting, standard methods were with SEM. used. Cells were lysed in the lysis buffer, and the Bradford method was precisely used to determine the Cell invasion and motility assay total protein content. SDS-PAGE separated 30 μg of Cell invasion was conducted in Matrigel (BD, protein under optimal conditions and blotted to a Franklin Lakes, NJ, USA)-coated Transwell inserts polyvinylidene difluoride membrane (Millipore). (6.5 mm, Costar, Manassas, VA, USA) comprising Specific primary antibodies have been tested for 8-μm pores polycarbonate filters as described above. membranes. Blots were washed and then blocked The inserts were covered with 50 μl of 1 mg/ml with respective secondary peroxidase-conjugated Matrigel matrix as it was recommended by the antibodies. Bands were visualized by chemi- manufacturer. 2×105 Cells in 200 μl of serum-free luminescence (Amersham Biosciences) as well. medium were plated in the upper chamber, whereas 600 μl of medium with 10% fetal bovine serum was Confocal Immunofluorescence microscopy added to the lower well. Cells that migrated to the By following the standard procedures, Cell lines lower surface of the membrane were fixed and stained were placed on culture slides (Costar, Manassas, VA, after 24 hours of incubation. Five random fields were USA). After 24 hrs, the cells were rinsed with counted at about ×10 magnification for each phosphate-buffered saline (PBS) and fixed with 4% membrane. The mean was calculated, and the paraformaldehyde in PBS. The cell membrane was information was displayed as mean ± s.d. from three permeabilized using 0.5% Triton X-100. These cells independent triplicate tests. Motility assays were were then blocked for 30 min in 10 % BSA in PBS and comparable to Matrigel's invasion assay except that then incubated with primary antibodies in 10 % BSA the Transwell insert was not covered with Matrigel. overnight at 4℃. After three washes in PBS, the slides were incubated for 1 hour in the dark with HOECHST 33342 staining, Flow cytometry FITC-conjugated secondary goat anti-mouse or goat analysis, and sorting of SP anti-rabbit antibodies (Invitrogen, Grand Island, NY, Cells were washed with PBS, detached from the USA). After three consecutive washes, the slides were culture plate with trypsin and EDTA, pelleted by stained with DAPI for 5 min to visualize the nuclei centrifugation, and resuspended in 37 °C DMEM and examined using a Carl Zeiss confocal imaging containing 2% FBS at 1×106 cells/ml. Cell staining has system (LSM 780) (Carl Zeiss, Jena, Germany). been conducted as earlier outlined. Either the cells were incubated alone, or in combination with known Clonogenic and sphere-formation assay ABC transporter inhibitor verapamil (50 μM, Sigma) For clonogenic assays, we plated cells at 1,000 at 5 μg/ml with Hoechst 33342 (Sigma, St, Louis, MO) cells / well in Matrigel (MG) or methylcellulose (MC) for 90 min at 37 °C., The cells were centrifuged and at the proportion of 1:1 in 100-200 μl and counted resuspended after staining in HBSS (Invitrogen, colonies 1-2 weeks after placing. Sphere-formation Carlsbad, CA) containing 1μg/ml propidium iodide culture has been conducted with slight changes. and maintained at 4 °C for flow cytometry evaluation Single-cell suspensions were plated at a distinct and sorting. Cell evaluation and sorting were density of feasible cells in 96-well plates of ultralow conducted on a 351- and 488-nm blue MoFlo attachment (Costar, Manassas, VA, USA). Cells were cytometer (Dako Cytomation, Fort Collins, CO) fitted cultivated in a serum-free mammary epithelial

http://www.thno.org Theranostics 2020, Vol. 10, Issue 21 9461 growth medium (MEGM), supplemented by 1:50 B27 All animal procedures were reviewed and approved (Invitrogen, Grand Island, NY, USA), 20 ng/ml by the Institutional Animal Care and Use Committee epidermal growth factor (EGF), 20 ng/ml basic of Shandong University. fibroblast growth factor (bFGF) (BD, Franklin Lakes, NJ, USA) and 10 μg/ml heparin (Sigma, St. Louis, Whole-mount staining MO, USA). Spheroid figures were counted after 7-10 At the indicated ages, the fourth inguinal glands days. Primary spheres were gathered for in-vitro were dissected and spread on a glass slide. After 2 propagation, dissociated into a single cell suspension hours of acidic alcohol fixation, the tissues were and plated 96-well sheets in ultralow connection. The hydrated and stained overnight in Carmine alum. secondary spheroid figures were counted after 14 Samples were then dehydrated, cleared, and mounted days of plating. by xylene. Biopsies immunohistochemistry Primary mouse mammary gland epithelial cell Fixed tissues were sectioned at a size of 4 μm. isolation Tissue pieces were stained according to the Preparation of the single-cell mammary gland manufacturer's protocol with Biotin-Streptavidin HRP was carried out as earlier outlined. Briefly, mammary Detection Systems. Gently, sections were de- thoracic and inguinal glands were dissected from paraffinized, and antigen repaired with sodium mice, the tissues were digested at 37 °C for 6-8 h in citrate and then incubated at room temperature with 3 DMEM/F12 supplemented with 10% FBS and 1% P/S % hydrogen peroxide for 10 minutes. Sections of the and 750 U/ml Collagenase and 250 U/ml tissue were blocked with goat serum at 37 °C for 15 hyaluronidase. After this step, the organoids were minutes and incubated at 4 °C overnight with the gathered by centrifugation then treated individually specified vaccine. Positive and negative staining with trypsin (0.5%) and dispase (5 mg/ml); regions were identified using Image-Pro Plus 5.1 Ammonium chloride was used for red blood cell lysis software (Media Cybernetics, Silver Spring, MD, after centrifugation. Unless otherwise mentioned, all USA). reagents were acquired from stem cell technology. In vivo tumorigenesis and metastasis assay Mouse mammary stem cell analysis A group of 6 Balb/c nude mice was injected Mammary glands were dissected from mice subcutaneously with infected cells into the left and aged seven weeks. Mammary stem cells were right flanks for tumorigenesis assay. The tumor size evaluated after mechanical dissociation. For isolation was assessed using calipers to measure tumor of stem/progenitor cells, the following antibodies dimensions for up to 42 days. Cells were resuspended were used: CD49f and CD24 (Stem cell technology, in PBS for metastasis assay, and the cell suspension eBioscience). Blocking was done for 10 min with rat was injected into nude mice's tail veins. All animals serum. Cells were stained for 30 min on ice and were maintained under the guidelines of Shandong washed with staining media. Then, cells were re- University and evaluated and approved by the suspended in staining media containing 7-amino- Institutional Animal Care and Use Committee actinomycin D (1 μg/ml) to stain dead cells. Cells (Shandong University, Jinan, China). Food and water were analyzed using an LSR II, Flow-jo, and sorted were supplied ad libitum. Mo flow cell sorter. SIRT4 knockout mice Establishment of SIRT4 stable expression and SIRT4 knockout (KO) mice were obtained from SIRT4, SIRT1 cell lines knockdown The Jackson Laboratory. All animals were numbered, PBabe.puro retroviral construct containing and experiments were conducted in a blinded fashion. human SIRT4 cDNA and pSuper.retro.puro with After data collection, genotypes were revealed, and human SIRT4 shRNA was prepared as previously animals assigned to groups for analysis. For treatment described [15]. The generation of retrovirus experiments, mice were randomized. None of the supernatants and the transfection of breast cancer mice with the appropriate genotype were excluded cells were carried out. Infected cells were chosen by from this study or used in any other experiments. adding 2 μg/ml puromycin to the 48-hour culture Mice had not undergone prior treatment or medium and then retained in a complete medium procedures. All mice were fed a standard chow diet with 0.5 μg/ml puromycin. The above protocols also ad libitum and housed in pathogen-free facility with generated empty retroviral-infected stable cell lines. standard controlled temperature, humidity, and light- shRNA against SIRT1 expressed in the pSuper vector dark cycle (12 h) conditions with no more than five was prepared as earlier outlined. The generation of mice per cage under the supervision of veterinarians. retrovirus supernatants and the transfection of breast

http://www.thno.org Theranostics 2020, Vol. 10, Issue 21 9462 cancer cells were chosen as outlined above, except for Chromatin Immunoprecipitation-qPCR infected cells by adding 400 μg/ml of G418. The Analysis expression of SIRT4 and SIRT1 was verified by the For ChIP analysis, 3.0 ml of purified chromatin study of qRT-PCR and Western blotting. was used with a Sybr green master mix (Applied Measurement of glutamine and glucose Biosystems, Foster City, CA, USA), and qPCR were analyzed using ABI 7900 thermocycler. Each response The BioProfile FLEX analyzer (Nova Biomedical) was conducted in triplicate, and the tests were was used to measure glutamine, ammonia, glucose, repeated at least twice in order to verify and lactate concentrations in culture media. Briefly, reproducibility. Values for the threshold cycle (Ct) new media were introduced to a 6-well cell plate, and were acquired for each gene or genomic region, and media metabolite concentrations were analyzed 6-9 information was analyzed using the standard curve hrs later and normalized to the number of cells in each technique. For ChIP-qPCR assessment, values were well. standardized to the input control. They expressed as a Chromatin immunoprecipitation and fold rise over-enrichment identified using IgG. The sequencing median expression ± S.E.M. has been recorded. ChIP protocol was adapted from the MagnaChIP Luciferase Reporter Assays kit protocol (Millipore). Cells were fixed with 1% Cells were transduced by BRCA1 reporter formaldehyde, lysed with Cell Lysis, and then (Signal BRCA1 Lentivirus Reporter, QIAGEN) and Nuclear Lysis buffers respecting concentration of chosen for Puro Resistance. The reporter activity was 2×107 cells per mL and finally sonicated (30-min cycle evaluated using the Dual-Glo Luciferase Assay on Covaris apparatus; KBioscience). Sheared System (Promega) and normalized to EF1 α-renilla chromatin was immunoprecipitated overnight using luciferase. the following antibodies: anti-H4K16ac, anti-H3K9ac, and rabbit IgG. 1/10 of the sheared chromatin was Microarray analysis used as a reference (Input). The immune complex The microarray studies were conducted by collection was realized with Protein G Sepharose Shanghai Biotechnology Corporation (Shanghai, (Sigma-Aldrich, P3296), 1h30 at +4°C. For GFP-ChIP, China). Total RNA was separated from three replicate GFP-Trap agarose beads were used (Chromotek). specimens of human podocytes using TRIzol reagent Rinses were done according to the MagnaChIP kit (Invitrogen). Agilent Whole Oligo protocol with low salt, high salt, and LiCL immune Microarray (Agilent Technologies, Santa Clara, CA, complex wash buffers. Finally, elution was performed USA) was used for transcriptome analysis. Microarray according to the IPure Kit protocol (Diagenode, Cat information was standardized using GeneSpring GX No C03010012) following the manufacturer's software (Agilent Technologies), and were instructions. Enriched DNA from ChIP and Input classified into numerous pathways using the David DNA fragments were end-repaired, extended with an database accessible on (https:/david.ncifcrf.gov/). ‘A’ base on the 3′end, ligated with indexed paired-end adaptors (NEXTflex, Bioo Scientific) using the Bravo Mass spectrometric metabolomics analysis Platform (Agilent), size- selected after 4 cycles of PCR Mouse mammary gland specimens were washed with AMPure XP beads (Beckman Coulter) and overnight in a speed vac and then homogenized in amplified by PCR for 10 cycles more. Fifty-cycle PBS (pH 7.4) using an Omni International bead single-end sequencings were performed using homogenizer device at 6.45 m/s for 30 sec. Illumina HiSeq 2000 (Illumina, San Diego, CA). Reads Homogenate was obtained using CHCl3: MeOH (2:1) were aligned to human genome hg19 with BWA aln to separate polar and nonpolar metabolites. Samples (v0.7.5a) and peak calling assessed using MACS 2.0 were centrifuged at 3000 rpm for 5 min and then with a q-value cut-off. Peak-calling analyses for each analyzed with LC-MS/MS using the selected reaction ChIP-seq were realized independently. Then we monitoring (SRM) method with positive/negative ion created one specific list of peaks for each mark, polarity switching to a Xevo TQ-S mass spectrometer. including familiar peaks in at least 2 replicates. The peak areas integrated using MassLynx 4.1 Annotation and motif analyses have been done with (Waters Inc.) have been normalized to the respective HOMER (v4.7.2), with a p-value <10-30. Integrative protein concentrations, and the resulting peak areas Genomics Viewer (IGV 2.3.34) was used for have been subjected to relative quantification analysis representation. Intensity profiles of ChIPseq data using Metaboanalyst 3.0 (www.metaboanalyst.ca). were done using ngsplot version 2.61.

http://www.thno.org Theranostics 2020, Vol. 10, Issue 21 9463

Analysis of TCGA breast cancer data immunohistochemistry staining also showed a By using the co-occurrence tool at cBioPortal for significant decrease in SIRT4 protein levels in breast Cancer Genomics, a TCGA breast cancer data set was tumor tissues (Figure S2). used to correlate the expression of SIRT4 and SIRT1 To assess the role of SIRT4 in mammary with an expression of 286 ribosomal processing genes. epithelium, whole-mount analyses were conducted WT P-values were obtained from Fisher's precise T-test. on 8-week-old SIRT4 wild-type (SIRT4 ) (n = 15) and -/- -/- The TCGA LADC dataset was also considered for an SIRT4 (n = 18) mice (Figure 1C). Noticeably, SIRT4 association of SIRT4 mRNA with disease-free and mice showed an increased ductal side-branching (p < overall survival was calculated using the cBioPortal 0.01, Figure 1D-F) and higher numbers of mammary hi hi neg survival analysis tool and the log-rank test to evaluate stem cells (MSCs; defined as CD24 CD49f in Lin statistical significance. gated cells) (p < 0.001, Figure 1G). Meanwhile, we found that the deletion of SIRT4 contributed to GSEA analysis terminal end bud (TEB) development in the intact Global mRNA expression profiles of breast mammary glands (Figure S3A). Importantly, limiting cancer samples from Gene Expression Omnibus dilution transplantation experiments with (GEO) were subject to Gene Set Enrichment Analysis LinnegCD24hiCD49fhi MSCs demonstrated that the (GSEA) using GSEA 2.0.9 software available on increase in lobuloalveolar structures in SIRT4-/- mice (http:/www.broadinstitute.org/gsea/) to reveal the directly relates to MSCs (Figure 1H-J). Together, these connection between SIRT4 expression and stem cell observations indicate that SIRT4 probably disrupts pathway signature. mammary gland development and that the deletion of SIRT4 strikingly promotes TEB development that Statistical analysis grows out from MSCs. The TCGA BC information was used to evaluate SIRT4 genomic copy modifications. TCGA SIRT4 deletion promotes mammary information analysis was performed using the Xena tumorigenesis method of USDSC, which is available on (http:// To evaluate the relative contribution of SIRT4 in xena.ucsc.edu). Statistical analysis was conducted mammary tumorigenesis, we then crossed SIRT4-/- using the statistical software program called SPSS mice with MMTV-Neu mice, which mimic the human (IBM. Armonk, New York, USA. Student two-tailed t luminal phenotype and develop spontaneous testing was also used to determine comparisons mammary tumors after acquiring either additional between distinct groups). Results are presented as mutations or epigenetic modifications [25, 26] Figure mean ±SD; probabilities less than 5% (p < 0.05) were 2F and Figure S4A showed the SIRT4 expression in considered to be statistically significant. crossed mice. The tumors developed in MMTV-neu; SIRT4-/- mice showed high tumor formation and lung Ethics Committee Approval and Patient metastasis compared to control models (Figure 2A-C; Consent Figure S4D). The overall survival rate of MMTV-neu; This study was reviewed and approved by the SIRT4-/- mice was significantly low compared with Ethics Committee of the Qilu Hospital of Shandong MMTV-Neu mice (Figure 2D). Besides, SIRT4 University and the Second Hospital of Shandong knockout significantly increased cell proliferation as University. measured by Ki67 expression and MSC population (Figure 2E; Figure S4B-C). Together, these results Results suggest that SIRT4 plays a key role in mammary SIRT4 expression is downregulated in breast tumor development, which might implicate MSC cancer and related to mammary gland formation. development and stemness SIRT4 inhibits self-renewal and expansion of Consistent with other studies [17, 24], our breast tumor-initiating cells (BTICs) analysis performed based on data from the TCGA To gain insight into the mechanism by which database validated that SIRT4 was downregulated in SIRT4 regulates mammary tumorigenesis, we breast cancer (Figure 1A; Figure S1A-C). Similarly, the performed RNA-seq with mammary cells from analysis of TCGA and 3 Gene Expression Omnibus SIRT4WT and SIRT4-/- mice (Figure 3A). A total of 3123 (GEO) datasets (GSE2034, 25307, 7390) showed that genes were found to be up-regulated, while 2477 decreased SIRT4 expression correlated with increased genes were negatively correlated with SIRT4 risk of disease progression and poor clinical outcome depletion (Figure 3B). The top 50 genes significantly (Figure 1B; Figure S1D). Besides, immunoblotting and differentially expressed were summarized in the

http://www.thno.org Theranostics 2020, Vol. 10, Issue 21 9464 heatmap (Figure S5). (GO)-analysis Consistently, the reduced SIRT4 expression was revealed the enriched signatures related to observed in the CSC-enriched populations, including endogenous SIRT4-dependent in breast mammospheres, side population (SP) cells, and cancer, namely, stem cell signaling (Figure 3C). EPCAM+ cells (Figure 3D-F).

Figure 1. SIRT4 expression is downregulated in breast cancer and related to mammary gland development and stemness. (A) SIRT4 expression levels in breast cancers compared to healthy tissues from the TCGA data set of 1247 samples. (B) Kaplan-Meier analysis indicating the overall survival of breast cancer patients with high (red) (n = 1944) or low (black) (n = 2007) SIRT4 expression. (C) SIRT4 KO mice were obtained from The Jackson Laboratory. SIRT4 knockout efficiency was measured by immunoblotting (left panel) and qRT- PCR (right panel). (D) Whole-mount analyses were conducted on 8-week-old SIRT4 wild-type (SIRT4WT) (n = 15), and SIRT4-/- (n = 18) mice

http://www.thno.org Theranostics 2020, Vol. 10, Issue 21 9465 and representative images of mammary gland side-branches are shown. (E) Percent of growth. (F) The number of mammary gland side-branches was quantified. (G) Distribution of Linneg mouse mammary cells according to their expression of CD24 and CD49f was analyzed on 8-week-old SIRT4-/- mice and littermate controls (left). Mammary stem cells (MSCs), according to their expression of CD24hiCD49fhi in Linneg (right), were quantified by a flow cytometric analysis. (H, I and J) Schematic representation of limiting dilution transplantation experiments with LinnegCD24hiCD49fhi MSCs (H). A total of 300, 1000, or 1 x 104 LinnegCD24hiCD49fhi MSCs isolated from 8-week-old SIRT4-/- mice and littermate controls were injected into the cleared fat pad of 3-week-old FVB/NJ female mice. Whole-mount analyses were then conducted at 6 weeks after injection (I, J). Representative images of mammary gland side-branches are shown (I). The resulting data were analyzed by the Chi-square test (p < 0.001) (J). Scale bars, 100 µm (D) and 50 µm (I).

Figure 2. SIRT4 deletion promotes mammary tumorigenesis. (A) SIRT4-/- mice were crossed with MMTV-Neu mice to generate MMTV-Neu transgenic mice in wild-type and SIRT4-null genetic backgrounds. The mammary tumor was developed at an average age of 240 days. Representative images are shown. Using these mice, we monitored (B) tumor weight, (C) tumor incidence, (D) overall survival time, (E) Ki67 and SIRT4 stained tumor sections, (F) SIRT4 protein expression in tumors isolated from MMTV-neu and MMTV-neu; SIRT4-/- mice. Data are means ± SEM. p < 0.01; t-test (B and C). Scale bars, 50 µm (E).

To further address the role of SIRT4 in cancer tumor-initiating cells. stemness, we next generated MDA-MB-231, BT549, Meanwhile, MTT, colony formation, and and SK-BR-3 cells overexpressing SIRT4, and MDA- bioluminescence imaging assay were performed to MB-468, MCF-7 as well as MCF-10A cells knocking measure the oncogenic phenotypes of SIRT4- down endogenous SIRT4 via lentiviral shRNA (Figure expressing MDA-MB-231 as well as SIRT4-deficient 3G-H; Figure S6; Figure S7A-D). As expected, the MCF-7 cells in vitro and in vivo (Figure S7E-J). In line ablation of SIRT4 caused an evident increase of with previous evidence, SIRT4 overexpression CD24-CD44+ populations, sphere formations, and SP disrupted the proliferation, colony formation, and cells, while forced expression of SIRT4 led to a xenograft tumor formation of MDA-MB-231 cells, noticeable reduction (Figure 3I-L). In order to test while ablation of SIRT4 caused the opposed effects on whether ectopic expression of SIRT4 could succeed in MCF-7 cells. Similarly, SIRT4 overexpression and altering the tumor-initiating frequency, we injected deficiency lead to a robust downregulation or up- transformed MDA-MB-231 cells with serial dilutions regulation of migration, mesenchymal markers, and into nude mice. None of the mice that were injected liver as well as lung metastasis, respectively (Figure with 103 SIRT4-expressing MDA-MB-231 cells formed S8). The findings mentioned above show that SIRT4 tumors (0 of 10 injected hosts), whereas 5 tumors plays a crucial role in the regulation of BTIC self- arose when 103 cells expressing a control vector were renewal and expansion and might protect mice from injected (Figure 3M). Thus, excessive expression of mammary tumorigenesis and lung metastasis by SIRT4 considerably decreased the number of breast blocking MSC formation.

http://www.thno.org Theranostics 2020, Vol. 10, Issue 21 9466

Figure 3. SIRT4 inhibits self-renewal and expansion of breast tumor-initiating cells (BTICs). (A) Heatmap summarizing genes differentially expressed in SIRT4-/- compared to SIRT4WT mice. (B) Volcano plot displaying differentially expressed genes. Up-regulated genes (3123) are highlighted in red. Down-regulated genes (2477) are highlighted in green. Black dots represent genes not differentially expressed. (C) Enrichment of a stem cell signaling in GSEA analysis of genes altered as described above. (D, E, F) Immunoblotting (upper panel) and mRNA expression (bottom panel) of SIRT4 in CSC-enriched mammospheres (D), side population (SP) cells (E), and EPCAM+ cells (F) as well as their corresponding controls, i.e., adherent cells (A), non-SP cells (C), and EPCAM- cells (E). (G) Quantification of CD44+/CD24- subpopulations (right panel) and immunoblotting of SIRT4 expression (left panel) in MDA-MB-231, MCF-7, and BT549 cells transfected with sh-SIRT4 or negative control (NC). (H) Quantification of CD44+/CD24- subpopulations (right panel) and immunoblotting of SIRT4 expression (left panel) in MDA-MB-468, MCF-10A, and SK-BR-3 cells transfected with SIRT4 or control vector (Vec). (I, J) Sphere formation efficiency of cells described in G (I) and H (J), respectively. (K, L) Hoechst SP assay of cells described in G (K) and H (L), respectively. (M) Tumor formation ability of MDA-MB-231 cells expressing control (Vector) or SIRT4 vector. The transfected MDA-MB-231 cells were assayed for the ability to form tumors by subaxillary injection of 1 × 106, 1 × 105, 10,000, 1,000, and 100 cells into nude mice. The numbers of tumors formed and the number of injections that were performed is listed for each population. Data are means ±SEM. **p < 0.01; t-test. Scale bars, 100 µm (I and J).

http://www.thno.org Theranostics 2020, Vol. 10, Issue 21 9467

transformed MDA-MB-231 cells indicated that overall Proteome-wide analysis of SIRT4 H4K16ac signals were lower and exhibited a short deficiency-induced expression in mouse peak on the region of BRCA1 in mammary mammary gland cells cells from SIRT4-/- mice and MDA-MB-231 cells To further identify the potential downstream transfected with control vectors (Figure 5C-D), targets of SIRT4, we performed the proteome-wide suggesting histone acetylation is associated with analysis with mammary epithelial cells from SIRT4WT activation of this promoter. To figure out whether this and SIRT4-/- mice (Figure 4A-B). As a result, a total of is the case, mammary cells from SIRT4WT and SIRT4-/- 176 proteins was significantly altered by SIRT4 mice, together with transformed MDA-MB-231, ablation. Among these proteins, 14 proteins were MCF-7, MDA-MB-468, and BT549 cells as described overlapped with the genes identified by RNA-seq above were chromatin immunoprecipitated (ChIP) for (Figure 4C top). Subsequent functional network IgG and H4K16ac, followed by qPCR, which analysis revealed the core position of SIRT1 (Figure validated the ChIP-seq result and demonstrated low 4C bottom), which was up-regulated by SIRT4 enrichment for H4K16ac -bound sequences at Exon 1 deficiency. Since SIRT1 has been reported to be rather than Exon 2 and 4 (Figure 5E-F; Figure S11B-D). involved in tumor initiation and progression as well Consistently, luciferase reporter assays in as cancer stemness, we hypothesized that SIRT1 transformed MDA-MB-231 and BT549 cells showed might be a potential target of SIRT4. that ectopic expression of SIRT4 activated the In agreement with the hypothesis, SIRT4 promoter of BRCA1, which was fully rescued by the deficiency increased SIRT1 expression at both mRNA catalytically inactive mutant of SIRT4 (H161Y) (Figure and protein levels (Figure 4D-E; Figure S9). In 5G), suggesting an essential role of its catalytical comparison, overexpression of SIRT4 caused reduced activity in regulating BRCA1 transcription. Similarly, SIRT1 expression, which was not observed in MDA- SIRT4 deficiency downregulated BRCA1 at both MB-231 and BT549 cells overexpressing catalytically mRNA and protein levels, whereas SIRT4 over- inactive SIRT4 H161Y mutation (Figure 4F). expression caused the opposite effect (Figure 5H-I; Consistently, the reverse relationship between SIRT4 Figure S12A-B). Additionally, the positive and SIRT1 proteins was further verified in human relationship between SIRT4 and BRCA1 expression mammary tissues by IHC staining (Figure S10). The was further verified by data from 3 GEO datasets above findings show that SIRT4 negatively regulates (GSE1456 and 2034) and in human mammary tissues SIRT1 expression in breast tumors. by IHC staining (Figure S12C-D). These data together provide compelling evidence that SIRT4 promotes SIRT4 deficiency down-regulates BRCA1 histone acetylation at the promoter region of BRCA1 expression to promote its transcription. SIRT1 has been reported to modulate the acetylation patterns of histones H3 and H4 and SIRT1 is required for SIRT4 deficiency-induced regulate both stemness and metastasis in breast cancer BCSC phenotype [27, 28]. Considering the inverse relationship between To prove the involvement of SIRT1 in SIRT4 and SIRT1, we next assessed the effect of SIRT4 SIRT4-deficiency-induced malignancy, we established on acetyl-histone H3 at lys9 (H3K9ac) and acetyl- stable breast cancer cell lines (MDA-MB-468 and histone H4 at lys16 (H4K16ac) as well as stem cell MCF-7) expressing two independent shRNAs that markers like Oct4, Sox2, and Nanog by immuno- target SIRT1 and SIRT4. As anticipated, blotting. As expected, SIRT4 deficiency increased the downregulation of [15] CD24-CD44+ populations were expression of these proteins. At the same time, observed in both cell lines (Figure 6A-B). Besides, the overexpression caused a reduction in the expression evaluation of stem cell markers further demonstrated of these proteins, and this was not observed in MDA- that the stemness phenotype favored by MB-231 and BT549 cells overexpressing catalytically SIRT4-deficiency was revoked by SIRT1 loss as well. inactive SIRT4 H161Y mutation (Figure 5A-B; Figure Coincident observations were achieved by S11A). immunoblotting and immunofluorescence that SIRT1 Next, we assessed genome-wide H4K16ac deficiency completely abrogated the alterations of binding by chromatin immunoprecipitation/high- SIRT4 loss on the potential downstream targets, throughput sequencing (ChIP-seq) analysis in H4K16ac and BRCA1 (Figure 6C-D; Figure S13A-B). mammary cells from SIRT4WT and SIRT4-/- mice and In vivo evidence from mouse xenografts injected with transformed MDA-MB-231 cells overexpressing the double knockdown cells revealed that the ablation SIRT4 or control vectors, respectively. As expected, of SIRT1 expression counteracted the effects of SIRT4 the analysis of enriched loci (peaks) in SIRT4-/- and decline on tumor growth (Figure 6E-G), which is

http://www.thno.org Theranostics 2020, Vol. 10, Issue 21 9468 consistent with the IHC data of Ki67 staining (Figure of SIRT1 by SIRT4 is vital for blocking the CSC S13C). These data altogether indicate that repression phenotype and tumor progression in breast cancer.

Figure 4. Proteome-wide analysis of SIRT4 deficiency-induced expression in mouse mammary gland cells. (A) Heatmap depicting proteins differentially expressed in mammary epithelial cells from SIRT4WT and SIRT4-/- mice. Up-regulated proteins are highlighted in pink. Down-regulated proteins are highlighted in purple. (B) Volcano plot displaying differentially expressed proteins in SIRT4-/- compared to SIRT4WT mice. (C) Venn diagram showing the overlap of genes (purple) and proteins (yellow)

http://www.thno.org Theranostics 2020, Vol. 10, Issue 21 9469 differentially expressed in mammary epithelial cells from SIRT4WT and SIRT4-/- mice (upper); Protein-Protein-Interaction Network including the 14 overlapped proteins in Venn diagram (bottom). (D) Representative IHC staining images of SIRT4 and SIRT1 in tumor sections isolated from SIRT4WT and SIRT4-/- mice. (E, F) Immunoblotting (upper panel) and mRNA expression (bottom panel) of SIRT1 in tumors isolated from SIRT4-/- and SIRT4WT mice (E), and in MDA-MB-231 and BT549 cells transfected with control (Vector), SIRT4 or mutated SIRT4 (H161Y) vector (F). Data are means ±SEM. **p < 0.01; t-test (E and F). Scale bars, 100 µm (D).

Figure 5. SIRT4 deficiency down-regulates BRCA1 expression. (A, B) Immunoblotting of acetyl-histone H4 at lys16 (H4K16ac), acetyl-histone H3 at lys9 (H3K9ac), Oct4, SOX2, and Nanog in mammary cells isolated from SIRT4-/- as well as SIRT4WT mice (A), and in MDA-MB-231 (left panel) as well as BT549 cells (right panel) transfected with control (Vector), SIRT4 or mutated SIRT4 (H161Y) vector (B). (C) Heatmap summarizing chromatin immunoprecipitation (ChIP)-seq data for H4K16ac, comparing mammary cells from SIRT4WT and SIRT4-/- mice (left panel), as well as MDA-MB-231 cells transfected with control (Vector) or SIRT4 vector (right panel). Profiles are centered on H4K16ac binding peaks and depict signal intensity (relative fold enrichment) with green color. (D) ChIP signal of IgG (left panel, red peaks), H3K9ac (middle, blue peaks), and H4K16ac (right panel, green peaks) in the indicated genomic regions of BRCA1 in cells described in C. (E, F) Mammary cells from SIRT4WT as well as SIRT4-/- mice (E) and transformed MDA-MB-468 cells (F) as described in C and Fig.3G were chromatin immunoprecipitated for IgG and H4K16ac. Pull-down at the putative H4K16ac binding sites was assessed by qRT- PCR and calculated as the percentage of IgG input. Error bars are SEM for 3 technical replicates. (G) Luciferase reporter assays showing the impact of SIRT4 deletion (left), as well as overexpression of SIRT4 and its mutation (H161Y) on BRCA1 promoters (right panel) in SIRT4-/- as well as SIRT4WT mammary cells, and MDA-MB-231 as well as BT549 cells, respectively. (H, I) Immunoblotting (H) and mRNA expression (I) of BRCA1 in cells described above. Data are means ±SEM. **p < 0.01; t-test.

http://www.thno.org Theranostics 2020, Vol. 10, Issue 21 9470

Figure 6. SIRT1 is required for the SIRT4 deficiency-induced BCSC phenotype. (A) Sphere formation efficiency of MDA-MB-468 (left) and MCF-7 cells (right panel) transfected with different combinations of control vector, sh-SIRT4, and sh-SIRT1. (B) Quantification of CD44+/CD24- subpopulations (C) SIRT4, SIRT1, H4K16ac, BRCA1, SOX2, Oct4, and Nanog in MDA-MB-468 (left) and MCF-7 cells (right panel) described above. (D) Immunofluorescence images of cells described in (A) stained with antibodies against BRCA1/DAPI. (E) Representative ventral view images of bioluminescence from mice described above. (F) Tumor volume. (G) Quantification of E. Data are means ±SEM. **p < 0.01; t-test. Scale bars, 100 µm (A) and 20 µm (D).

http://www.thno.org Theranostics 2020, Vol. 10, Issue 21 9471

and CD24-CD44+ populations induced by SIRT4 Glutamine metabolism disorder mediates depletion were also abrogated by GLS1 inhibition SIRT4-induced SIRT1 inhibition in breast (Figure 7E-G; Figure S15B-D). Specifically, over- cancer cell expression of SIRT4 in MDA-MB-231 and BT549 cells Our previous studies have clarified the tumor- triggered phosphorylation of AMPKα, which in turn suppressive role of SIRT4 in HCC via mediating inhibited SIRT1. In contrast, AMPKα knockdown glutaminolysis. To examine whether SIRT4 exerts almost completely recovered the expression of SIRT1, similar functions in breast cancer, we conducted suggesting the essential role of AMPKα in SIRT4- LC-MS/MS analysis of metabolites in mammary modulated expression of SIRT1 (Figure S15E-F). In epithelial cells from SIRT4WT and SIRT4-/- mice (n = 6). vivo, evidence showed similar results that BPTES Consistently, glutamine was significantly up- treatment successfully reversed the tumor-promoting regulated by SIRT4 deficiency (Figure 7A). effects by SIRT4 deficiency (Figure 7H-J). Together Moreover, SIRT4-depleted mouse mammary these results signified that SIRT4-deletion enhanced gland cells exhibited significantly increased metabolic flux into glutaminolysis of mammary cells glutamine uptake and NH4+ production (Figure 7B). in conjunction with phosphorylation of AMPKα and Importantly, glucose uptake and lactate production the subsequent inhibition of SIRT1 and its were not affected, indicating that SIRT4 loss increased downstream targets. the ability of cells to utilize glutamine for mitochondrial energy production, as addressed by EX-527 significantly eliminates SIRT4 other studies [17]. Similar observations were detected depletion-induced BTICs and xenograft in SIRT4-knockdown human MDA-MB-468 and formation MCF-7 cells (Figure S14A-B). Conversely, forced EX-527 or Selisistat, a highly potent and selective expression of wild-type (but not catalytic H161Y inhibitor of SIRT1, used to being developed as a mutant) SIRT4 in MDA-MB-231 and BT549 cells led to disease-modifying therapy for Huntington's Disease decreases in glutamine uptake and NH4 + [34]. With the accumulated evidence about SIRT1's production. However, it was not observed in those function in cancer, EX-527 and other SIRT inhibitors cells overexpressing catalytically inactive SIRT4 have been studied as potential therapies in multiple H161Y mutation (Figure S14C-D). Collectively, these cancer cells, including melanoma [35], hepatocellular data indicated that SIRT4 depletion affected carcinoma [36], chronic lymphocytic leukemia [37], glutaminolysis in mammary tumor cells. and lung cancer [38]. Despite an inhibitory or In our previous study, we found that as a synergistic suppressive effect observed in the cells, in mitochondrial sirtuin, SIRT4 could exert its tumor- vivo confirmative observations are missing. suppressive function in HCC by inhibiting glutamine Concerning breast cancer, EX-527 has been implicated metabolism and thereby regulating AMP-activated in chemoresistance in several studies when being protein kinase (AMPK) /mTOR Axis [15]. AMPK, a combined with different drugs [39-41]. highly conserved sensor of cellular energy and To explore the potential clinical significance of nutrient status, has been reported to phosphorylate EX-527, MDA-MB-468, and MCF-7, cells were treated SIRT1 and inhibit its deacetylation activity in lung with or without EX-527 at 38 nM dose. Not cancer and HCC [29, 30]. Given the negative surprisingly, EX-527 treatment eliminated the regulation of SIRT4 on SIRT1 and the intricate increase of SP cells, mammosphere formation, and relationship between AMPK and SIRT1, we CD24-CD44+ populations induced by shSIRT4 in both hypothesized that glutamine metabolism disorder cell lines (Figure 8A-C). Coincident observations were might be involved in SIRT4-induced SIRT1 inhibition. achieved by immunoblotting and immuno- To test this hypothesis, we treated the cells with fluorescence assay that SIRT1 inhibition completely bis-2-(5-phenylacetoamido-1,2,4-thiadiazol-2-yl)ethyl abrogated the alterations of SIRT4 loss on the sulfide (BPTES) or 968, inhibitors of 1 potential downstream targets, H4K16ac, BRCA1, and (GLS1), which is the first studied necessary SOX2 (Figure 8D). In vivo evidence from mouse for mitochondrial glutamine metabolism and its xenografts injected with transformed cells with or inhibition limits entry of glutamine flux into the TCA without EX-527 treatment revealed that SIRT1 cycle [31-33]. Treatment of BPTES or 968 successfully inhibition counteracted the effects of SIRT4 decline on reversed the altered protein levels by SIRT4 tumor growth (Figure 8E-F). Collectively, these data deficiency, including SIRT1 and its downstream suggest that EX-527 might be an option for breast H4K16ac and BRCA1 as well as SOX2 (Figure 7C-D; cancer patients with low SIRT4 expression. Figure S15A). Likewise, SP cells, sphere formations,

http://www.thno.org Theranostics 2020, Vol. 10, Issue 21 9472

Figure 7. Glutamine metabolism disorder mediates SIRT4-induced SIRT1 inhibition in breast cancer cells. (A) Heatmap showing the changes in metabolites levels between SIRT4WT and SIRT4-/- mice. Up-regulated metabolites are highlighted in red. Down-regulated metabolites are highlighted in purple. (B) Measurement of Glutamine uptake (upper left), NH4+ production (upper right), Glucose uptake (bottom left), and lactate production (bottom right panel) in mammary epithelial cells from SIRT4WT and SIRT4-/- mice. (C, D) Immunoblotting of indicated proteins isolated from SIRT4WT and SIRT4-/-mammary cells with or without BPTES (10 µM) (C) and 968 (10 µM) (D) treatment. (E, F, G) Quantification of Hoechst SP assay (E), sphere formation efficiency (F), and CD44+/CD24- subpopulations (G) in SIRT4WT and SIRT4-/-mammary cells with or without BPTES (left) and 968 (right panel) treatment. (H, I) Representative ventral view images of bioluminescence from mice with injections of cells described above (H) and its quantification (I). (J) Representative IHC staining images of BRCA1 in tumor sections isolated from mice described in H. Data are means ± SEM. **p < 0.01; t-test. Scale bars, 100 µm (J).

http://www.thno.org Theranostics 2020, Vol. 10, Issue 21 9473

Figure 8. EX-527 significantly eliminates SIRT4 depletion-induced BTICs and xenograft formation. (A, B, C) Quantification of Hoechst SP assay (A), sphere formation efficiency (B), and CD44+/CD24- subpopulations (C) in transformed MDA-MB-468 (left) and MCF-7 cells (right panel) described in Fig. 3G with or without treatment of EX-527, a highly potent and selective inhibitor of SIRT1. (D) Immunoblotting of indicated proteins isolated from transformed MDA-MB-468 (left) and MCF-7 cells (right panel) described above with or without EX-527 treatment. (E, F) Representative ventral view images of bioluminescence from mice with injections of cells described above (E) and its quantification (F). (G) A schematic model was illustrating the biological processes regulated by SIRT4 in breast cancer. Data are means ± SEM. p < 0.01; t-test.

http://www.thno.org Theranostics 2020, Vol. 10, Issue 21 9474

recognition sequences on SIRT3 promoter [52]. Discussion Besides, SIRT1 was shown to regulate SIRT3 The studies on SIRT4 have been increasingly expression after oxygen and glucose deprivation reported during the past few years; however, SIRT4 through inhibiting the AMPK-PGC1 pathway, which remains to be the least well-characterized sirtuin. The might be an essential modulator in blood-brain mitochondrial SIRT4 is intimately involved in barrier physiology [53]. To efficiently deacetylate metabolic regulation with the inhibitory effect against targets such as , H3K9, and H4K16, SIRT1 requires secretion in pancreatic β cells or fatty acid autodeacetylation to be activated, which could be oxidation in liver and muscle cells [42, 43]. Consistent restricted by SIRT7 during adipogenesis in mice [54]. with the importance of mitochondrial metabolism in The protein-protein interaction network of the human tumorigenesis, SIRT4 was found to mainly exert sirtuin family showed that SIRT4 was interacting with tumor-suppressive function under its metabolic all the sirtuin family, including SIRT1 [55], which is regulatory role [44]. SIRT4-mediated inhibition of also confirmed by our proteome-wide analysis of glutamine catabolism has been implicated in DNA SIRT4 deficiency-induced expression in mouse damage response, cell proliferation, migration, and mammary gland cells. Our data illustrate that SIRT4 invasion in multiple cancers [15, 17, 45]. In terms of negatively regulates SIRT1 expression in breast breast cancer, low SIRT4 expression was linked with tumors. poor overall survival in breast cancer patients, Mechanistically, SIRT4 was known to attenuate especially in Luminal A patients [21]. Consistent with the activation effect of SIRT1 on PPARα transcription this finding, we also revealed that the correlation by interfering with the binding of SIRT1 with PPARα between decreased SIRT4 expression and increased response elements and thus represses hepatic fatty risk of disease progression. To the best of our acid oxidation [56]. In this study, we demonstrate that knowledge, no direct role of SIRT4 in regulating SIRT4 activity dampens breast cancer stemness by reprogramming or stemness has been reported. modulating SIRT1 expression, as supported by the Correctly, SIRT4 has been shown to prevent the evidence that genetic and pharmacological inhibition senescence of somatic cells, which might facilitate the of SIRT1 expression or activity could successfully reprogramming [46, 47]. Here, with SIRT4-/- mice and eradicate the expansion of BTICs or xenograft crossed MMTV-neu; SIRT4-/- mice, we identified an formation induced by SIRT4 loss. More importantly, unrecognized role of SIRT4 in regulating mammary the metabolomic analysis identified glutamine gland development and stemness, contributing to metabolism disorder in mammary gland cells from mammary tumorigenesis. This result was further SIRT4-/- mice, which is consistent with the metabolic validated by the RNA-seq and GO analysis of genes regulatory role of SIRT4 in glutamine uptake and significantly differentially expressed in response to utilization. GLS1 inhibitors, BPTES, or 968 blunts the SIRT4 knockout. Breast tumor initiating cells can be xenograft formation and increase in SIRT1 protein enriched by sorting for CD24-CD44+ cells, by selecting levels as well as the BTIC phenotype observed in for side-population cells that efflux Hoechst dyes, or SIRT4 null cells. Our findings thus highlight the by isolating mammospheres from suspension cultures unique effect of SIRT4 on suppressing the expression [48-50]. Consistently, ablation of SIRT4 caused an and function of SIRT1 through modulating glutamine evident expansion while SIRT4 overexpression metabolism, providing a novel cross-talk between inhibited self-renewal ability of these BTICs. mitochondrial and nuclear sirtuins. The complex system of interactions mediated by As the most commonly diagnosed cancer in mammalian sirtuins is mostly involved in many women worldwide, breast cancer is a multifactorial metabolic processes as well as in different diseases. genetic disease. 85 to 90% of breast tumors are The cellular distribution and biological functions of sporadic, which are uniquely characterized by an these proteins are different though; different sirtuins altered epigenome. Deregulated histone epigenome, also share a conserved catalytic domain and control along with other epigenetic alterations play a vital similar cellular processes, suggesting a co-ordinated role in the initiation and progression of breast cancer. mode of action [47, 51]. While there is a substantial Global loss of H4K16ac and under-expressed H3K9ac knowledge of the biological functions of sirtuins, is a hallmark of breast cancer, the expression pattern studies about the interplay and potential cross- of which could be modulated by SIRT1 through direct regulation within this family are still limited. deacetylation [27, 57]. Moreover, the colocalization of Recently, the SIRT1-SIRT3 axis has been identified to SIRT1 and its H3 acetylated targets was identified on regulate the cellular response to oxidative stress and the BRCA1 promoter in a subtype-specific manner etoposide. SIRT1-silencing increases SIRT3 promoter [27]. So far, no direct evidence has linked SIRT4 to activity depending on the presence of SP1 and ZF5 modulation of H4K16ac and BRCA1 expression. Until

http://www.thno.org Theranostics 2020, Vol. 10, Issue 21 9475 very recently, the first SANT domain (SANT1) of Author Contributions BRCA1 was reported to be a histone binding domain CX.W and YS.W conceived and supervised the with specificity for the histone H4 N-terminal tail. project, acquired funding, performed and designed Acetylation of H4K16 consequently abolishes binding the majority of experiments. LT.D and PM wrote the of SANT1 to H4 [58], suggesting potential regulation manuscript. LT. D, QL.J, JL, PL.L, FW, and YD.R of BRCA1 expression by H4K16ac. Consistently, our performed glucose experiments and supported the ChIP-seq results uncovered that SIRT4 deficiency biochemical assays. down-regulates BRCA1 expression through modulation of H4K16ac rather than the well-defined Competing Interests H3K9ac, which is further confirmed by following The authors have declared that no competing ChIP-qPCR assay. Furthermore, we also validated the interest exists. two putative H4K16ac binding sites on Exon 1 of BRCA1, which has not been reported before. References 1. Kosciuk T, Wang M, Hong JY, Lin H. Updates on the epigenetic roles of Conclusions sirtuins. Curr Opin Chem Biol. 2019; 51: 18-29. 2. O'Callaghan C, Vassilopoulos A. Sirtuins at the crossroads of stemness, aging, In conclusion, we found that SIRT4 regulation of and cancer. Aging Cell. 2017; 16: 1208-1218. 3. Zhu S, Dong Z, Ke X, Hou J, Zhao E, Zhang K, et al. The roles of sirtuins family BITCs negatively modulated mammary in cell metabolism during tumor development. Semin Cancer Biol. 2018. tumorigenesis through suppression of SIRT1 via 4. Tanno M, Sakamoto J, Miura T, Shimamoto K, Horio Y. Nucleocytoplasmic shuttling of the NAD+-dependent SIRT1. J Biol Chem. glutamine metabolism and that SIRT4 deficiency 2007; 282: 6823-6832. decreased H4K16ac and BRCA1 expression in breast 5. Frazzi R. SIRT1 in Secretory Organ Cancer. Front Endocrinol (Lausanne). 2018; 9: 569. cancer (Figure 8G). These findings shed light on novel 6. Hong WG, Pyo JS. The clinicopathological significance of SIRT1 expression in molecular mechanisms of glutamine metabolism colon cancer: An immunohistochemical study and meta-analysis. Pathol Res Pract. 2018; 214: 1550-1555. modulation by SIRT4 and cross-talk between 7. Karbasforooshan H, Roohbakhsh A, Karimi G. SIRT1 and : The mitochondrial and nuclear sirtuins, suggesting that role in breast, lung and prostate cancers. Exp Cell Res. 2018; 367: 1-6. 8. Yang Q, Wang B, Gao W, Huang S, Liu Z, Li W, et al. SIRT1 is downregulated SIRT4 has tumor-suppressive activity and may serve in gastric cancer and leads to G1-phase arrest via NF-kappaB/Cyclin D1 as a novel therapeutic target in breast cancer. signaling. Mol Cancer Res. 2013; 11: 1497-1507. 9. Zhang S, Huang S, Deng C, Cao Y, Yang J, Chen G, et al. Co-ordinated overexpression of SIRT1 and STAT3 is associated with poor survival outcome Abbreviations in gastric cancer patients. Oncotarget. 2017; 8: 18848-18860. 10. Mvunta DH, Miyamoto T, Asaka R, Yamada Y, Ando H, Higuchi S, et al. AMPK: AMP-activated protein kinase; bFGF: Overexpression of SIRT1 is Associated With Poor Outcomes in Patients With Ovarian Carcinoma. Appl Immunohistochem Mol Morphol. 2017; 25: 415-421. Basic fibroblast growth factor; BPTES: Bis-2-(5- 11. Wang C, Yang W, Dong F, Guo Y, Tan J, Ruan S, et al. The prognostic role of phenylacetoamido-1,2,4-thiadiazol-2-yl)ethyl sulfide; Sirt1 expression in solid malignancies: a meta-analysis. Oncotarget. 2017; 8: 66343-66351. BTICs: Breast tumor-initiating cells; ChIP-seq: 12. Li S, Zheng W. Mammalian Sirtuins SIRT4 and SIRT7. Prog Mol Biol Transl Chromatin immunoprecipitation/high-throughput Sci. 2018; 154: 147-168. 13. Min Z, Gao J, Yu Y. The Roles of Mitochondrial SIRT4 in Cellular Metabolism. sequencing; EGF: Epidermal growth factor; GEO: Front Endocrinol (Lausanne). 2018; 9: 783. Gene expression omnibus; GLS1: Glutaminase 1; 14. Hu Y, Lin J, Lin Y, Chen X, Zhu G, Huang G. Overexpression of SIRT4 inhibits the proliferation of gastric cancer cells through cell cycle arrest. Oncol Lett. GSEA: Gene set enrichment analysis; HDACs: 2019; 17: 2171-2176. Histone deacetylases; MC: methylcellulose; MEGM: 15. Wang YS, Du L, Liang X, Meng P, Bi L, Wang YL, et al. Sirtuin 4 Depletion Promotes Hepatocellular Carcinoma Tumorigenesis Through Regulating Mammary epithelial growth medium; MG: Matrigel; Adenosine-Monophosphate-Activated Protein Kinase Alpha/Mammalian Target of Rapamycin Axis in Mice. Hepatology. 2019; 69: 1614-1631. MSC: Mammary stem cells; NAD+: Nicotinamide 16. Fu L, Dong Q, He J, Wang X, Xing J, Wang E, et al. SIRT4 inhibits malignancy adenine dinucleotide; SIR2: Silent information progression of NSCLCs, through mitochondrial dynamics mediated by the ERK-Drp1 pathway. Oncogene. 2017; 36: 2724-2736. regulator 2; SP: Side population; SRM: Selected 17. Jeong SM, Xiao C, Finley LW, Lahusen T, Souza AL, Pierce K, et al. SIRT4 has reaction monitoring; TCGA: The cancer genome atlas; tumor-suppressive activity and regulates the cellular metabolic response to DNA damage by inhibiting mitochondrial glutamine metabolism. Cancer Cell. TEB: Terminal end bud; TICs: Tumor-initiating cells. 2013; 23: 450-463. 18. Csibi A, Fendt SM, Li C, Poulogiannis G, Choo AY, Chapski DJ, et al. The mTORC1 pathway stimulates glutamine metabolism and cell proliferation by Supplementary Material repressing SIRT4. Cell. 2013; 153: 840-854. Supplementary figures. 19. Huang G, Cheng J, Yu F, Liu X, Yuan C, Liu C, et al. Clinical and therapeutic significance of sirtuin-4 expression in colorectal cancer. Oncol Rep. 2016; 35: http://www.thno.org/v10p9458s1.pdf 2801-2810. 20. Nakahara Y, Yamasaki M, Sawada G, Miyazaki Y, Makino T, Takahashi T, et al. Downregulation of SIRT4 Expression Is Associated with Poor Prognosis in Acknowledgments Esophageal Squamous Cell Carcinoma. Oncology. 2016; 90: 347-355. 21. Shi Q, Liu T, Zhang X, Geng J, He X, Nu M, et al. Decreased sirtuin 4 We thank Haiyang Guo for critical comments expression is associated with poor prognosis in patients with invasive breast and discussion on this paper. We thank Ying Han and cancer. Oncol Lett. 2016; 12: 2606-2612. 22. Molyneux G, Geyer FC, Magnay FA, McCarthy A, Kendrick H, Natrajan R, et Wen Jiang for supporting the surgical al. BRCA1 basal-like breast cancers originate from luminal epithelial instrumentation of animals' studies, mammalian cell progenitors and not from basal stem cells. Cell Stem Cell. 2010; 7: 403-417. 23. Smalley MJ, Reis-Filho JS, Ashworth A. BRCA1 and stem cells: tumour culture, and biochemical assays. typecasting. Nat Cell Biol. 2008; 10: 377-379.

http://www.thno.org Theranostics 2020, Vol. 10, Issue 21 9476

24. Igci M, Kalender ME, Borazan E, Bozgeyik I, Bayraktar R, Bozgeyik E, et al. 51. Lin S, Xing H, Zang T, Ruan X, Wo L, He M. Sirtuins in mitochondrial stress: High-throughput screening of Sirtuin family of genes in breast cancer. Gene. Indispensable helpers behind the scenes. Ageing Res Rev. 2018; 44: 22-32. 2016; 586: 123-128. 52. Carnevale I, Pellegrini L, D'Aquila P, Saladini S, Lococo E, Polletta L, et al. 25. Guy CT, Webster MA, Schaller M, Parsons TJ, Cardiff RD, Muller WJ. SIRT1-SIRT3 Axis Regulates Cellular Response to Oxidative Stress and Expression of the neu protooncogene in the mammary epithelium of Etoposide. J Cell Physiol. 2017; 232: 1835-1844. transgenic mice induces metastatic disease. Proc Natl Acad Sci U S A. 1992; 89: 53. Chen T, Dai SH, Li X, Luo P, Zhu J, Wang YH, et al. Sirt1-Sirt3 axis regulates 10578-10582. human blood-brain barrier permeability in response to ischemia. Redox Biol. 26. Herschkowitz JI, Simin K, Weigman VJ, Mikaelian I, Usary J, Hu Z, et al. 2018; 14: 229-236. Identification of conserved gene expression features between murine 54. Fang J, Ianni A, Smolka C, Vakhrusheva O, Nolte H, Kruger M, et al. Sirt7 mammary carcinoma models and human breast tumors. Genome Biol. 2007; 8: promotes adipogenesis in the mouse by inhibiting autocatalytic activation of R76. Sirt1. Proc Natl Acad Sci U S A. 2017; 114: E8352-E8361. 27. Rifai K, Judes G, Idrissou M, Daures M, Bignon YJ, Penault-Llorca F, et al. 55. Sharma A, Costantini S, Colonna G. The protein-protein interaction network SIRT1-dependent epigenetic regulation of H3 and H4 histone acetylation in of the human Sirtuin family. Biochim Biophys Acta. 2013; 1834: 1998-2009. human breast cancer. Oncotarget. 2018; 9: 30661-30678. 56. Laurent G, de Boer VC, Finley LW, Sweeney M, Lu H, Schug TT, et al. SIRT4 28. Shi L, Tang X, Qian M, Liu Z, Meng F, Fu L, et al. A SIRT1-centered circuitry represses peroxisome proliferator-activated receptor alpha activity to suppress regulates breast cancer stemness and metastasis. Oncogene. 2018; 37: hepatic fat oxidation. Mol Cell Biol. 2013; 33: 4552-4561. 6299-6315. 57. Elsheikh SE, Green AR, Rakha EA, Powe DG, Ahmed RA, Collins HM, et al. 29. Zhang Y, Sun C, Xiao G, Shan H, Tang L, Yi Y, et al. S-nitrosylation of the Global histone modifications in breast cancer correlate with tumor Peroxiredoxin-2 promotes S-nitrosoglutathione-mediated lung cancer cells phenotypes, prognostic factors, and patient outcome. Cancer Res. 2009; 69: apoptosis via AMPK-SIRT1 pathway. Cell Death Dis. 2019; 10: 329. 3802-3809. 30. Zhou X, Chen J, Chen L, Feng X, Liu Z, Hu L, et al. Negative regulation of 58. Weaver TM, Liu J, Connelly KE, Coble C, Varzavand K, Dykhuizen EC, et al. Sirtuin 1 by AMP-activated protein kinase promotes -induced The EZH2 SANT1 domain is a histone reader providing sensitivity to the senescence in hepatocellular carcinoma xenografts. Cancer Lett. 2017; 411: modification state of the H4 tail. Sci Rep. 2019; 9: 987. 1-11. 31. Curthoys NP, Watford M. Regulation of glutaminase activity and glutamine metabolism. Annu Rev Nutr. 1995; 15: 133-159. 32. Wang JB, Erickson JW, Fuji R, Ramachandran S, Gao P, Dinavahi R, et al. Targeting mitochondrial glutaminase activity inhibits oncogenic transformation. Cancer Cell. 2010; 18: 207-219. 33. Robinson MM, McBryant SJ, Tsukamoto T, Rojas C, Ferraris DV, Hamilton SK, et al. Novel mechanism of inhibition of rat kidney-type glutaminase by bis-2-(5-phenylacetamido-1,2,4-thiadiazol-2-yl)ethyl sulfide (BPTES). Biochem J. 2007; 406: 407-414. 34. Westerberg G, Chiesa JA, Andersen CA, Diamanti D, Magnoni L, Pollio G, et al. Safety, pharmacokinetics, pharmacogenomics and QT concentration-effect modelling of the SirT1 inhibitor selisistat in healthy volunteers. Br J Clin Pharmacol. 2015; 79: 477-491. 35. Wilking MJ, Singh CK, Nihal M, Ndiaye MA, Ahmad N. Sirtuin deacetylases: a new target for melanoma management. Cell Cycle. 2014; 13: 2821-2826. 36. Ceballos MP, Decandido G, Quiroga AD, Comanzo CG, Livore VI, Lorenzetti F, et al. Inhibition of sirtuins 1 and 2 impairs cell survival and migration and modulates the expression of P-glycoprotein and MRP3 in hepatocellular carcinoma cell lines. Toxicol Lett. 2018; 289: 63-74. 37. Bhalla S, Gordon LI. Functional characterization of NAD dependent de-acetylases SIRT1 and SIRT2 in B-Cell Chronic Lymphocytic Leukemia (CLL). Cancer Biol Ther. 2016; 17: 300-309. 38. Chen G, Zhang B, Xu H, Sun Y, Shi Y, Luo Y, et al. Suppression of Sirt1 sensitizes lung cancer cells to WEE1 inhibitor MK-1775-induced DNA damage and apoptosis. Oncogene. 2017; 36: 6863-6872. 39. Wang Y, Zeng J, Wu W, Xie S, Yu H, Li G, et al. Nicotinamide N-methyltransferase enhances chemoresistance in breast cancer through SIRT1 protein stabilization. Breast Cancer Res. 2019; 21: 64. 40. Kim HB, Lee SH, Um JH, Oh WK, Kim DW, Kang CD, et al. Sensitization of multidrug-resistant human cancer cells to Hsp90 inhibitors by down-regulation of SIRT1. Oncotarget. 2015; 6: 36202-36218. 41. Fatehi D, Soltani A, Ghatrehsamani M. SRT1720, a potential sensitizer for radiotherapy and cytotoxicity effects of NVB-BEZ235 in metastatic breast cancer cells. Pathol Res Pract. 2018; 214: 889-895. 42. Argmann C, Auwerx J. Insulin secretion: SIRT4 gets in on the act. Cell. 2006; 126: 837-839. 43. Nasrin N, Wu X, Fortier E, Feng Y, Bare OC, Chen S, et al. SIRT4 regulates fatty acid oxidation and mitochondrial gene expression in liver and muscle cells. J Biol Chem. 2010; 285: 31995-32002. 44. Zhu Y, Yan Y, Principe DR, Zou X, Vassilopoulos A, Gius D. SIRT3 and SIRT4 are mitochondrial tumor suppressor proteins that connect mitochondrial metabolism and carcinogenesis. Cancer Metab. 2014; 2: 15. 45. Miyo M, Yamamoto H, Konno M, Colvin H, Nishida N, Koseki J, et al. Tumour-suppressive function of SIRT4 in human colorectal cancer. Br J Cancer. 2015; 113: 492-499. 46. Castex J, Willmann D, Kanouni T, Arrigoni L, Li Y, Friedrich M, et al. Inactivation of Lsd1 triggers senescence in trophoblast stem cells by induction of Sirt4. Cell Death Dis. 2017; 8: e2631. 47. Hsu YC, Wu YT, Tsai CL, Wei YH. Current understanding and future perspectives of the roles of sirtuins in the reprogramming and differentiation of pluripotent stem cells. Exp Biol Med (Maywood). 2018; 243: 563-575. 48. Al-Hajj M, Wicha MS, Benito-Hernandez A, Morrison SJ, Clarke MF. Prospective identification of tumorigenic breast cancer cells. Proc Natl Acad Sci U S A. 2003; 100: 3983-3988. 49. Patrawala L, Calhoun T, Schneider-Broussard R, Zhou J, Claypool K, Tang DG. Side population is enriched in tumorigenic, stem-like cancer cells, whereas ABCG2+ and ABCG2- cancer cells are similarly tumorigenic. Cancer Res. 2005; 65: 6207-6219. 50. Ponti D, Costa A, Zaffaroni N, Pratesi G, Petrangolini G, Coradini D, et al. Isolation and in vitro propagation of tumorigenic breast cancer cells with stem/progenitor cell properties. Cancer Res. 2005; 65: 5506-5511.

http://www.thno.org